Large Language Models

Latest papers

• Layer-wise Positional Bias in Short-Context Language Modeling

  Maryam Rahimi, Mahdi Nouri, Yadollah Yaghoobzadeh · 7 August 2026

  Transformer language models systematically prefer tokens at specific input positions regardless of semantic relevance---a phenomenon known as positional bias. Prior work characterizes this bias in model behavior through performance drops in long-context tasks or in model architecture through attenti…

• When Do Prompt-Side Agent Playbooks Transfer? Accuracy, Cost, and Runtime Shift in Agent Deployment

  Weihong Lin, Lin Sun, Xiangzheng Zhang · 7 August 2026

  Prompt-side playbooks can improve tool-using language agents without retraining, but their portability beyond the source setting is unclear. We study frozen playbook transfer under a shared distill--validate--transfer protocol. On ALFWorld, transfer is beneficial under controlled greedy decoding and…

• SemiAdapt-Instruct: Extensible Instruction Tuning via Latent Domain-Specialised Adapters

  Josh McGiff, Salma Mekaoui, Robert Shanahan, Nikola S. Nikolov · 7 August 2026

  Instruction-tuned LLMs are deployed into environments where domains evolve, yet extending a fine-tuned model's capabilities without full retraining remains an unsolved practical challenge. We present SemiAdapt-Instruct, a modular framework that discovers latent instruction domains, trains per-domain…

• On-Policy Self-Distillation without Any Supervision

  Yijiang Li, Bingyang Wang, Yijun Liang, Yunjie Tian, Di Fu, Nuno Vasconcelos · 7 August 2026

  On-policy (Self-)Distillation (OPD / OPSD) has shown strong potential for post-training large language models (LLMs). However, existing methods still rely heavily on external supervision, including ground-truth signals, environmental feedback, or guidance from larger models, and therefore fall short…

• Position: It's Time to Optimize LLMs for Self-Consistency

  Itamar Pres, Belinda Z. Li, Laura Ruis, Zifan Carl Guo, Keya Hu, Mehul Damani, Isha Puri, Ekdeep Singh Lubana, Jacob Andreas · 7 August 2026

  Despite ever-increasing sophistication in language model (LM) pre- and post-training pipelines, many important failures persist: models overcondition on user framing ("sycophancy"), exhibit incomplete logical generalization, and produce confident but incorrect responses. We argue that these failures…

• Cautious Context Steering for Language Model Personalization

  Gihoon Kim, Jeyoung Lee, Suhan Woo, Sekwon Oh, Minsu Jeon, Hyounsoo Han, Euntai Kim · 7 August 2026

  Personalizing language models (LMs) to individual user preferences is essential for aligning responses with diverse goals and backgrounds. Existing methods typically train a separate adapter for each user or learn a reward model whose scores depend on the user. Despite explicitly optimizing for each…

• DASH: Divergence-Adaptive Supervision Horizons for On-Policy Self-Distillation of Reasoning Models

  ZhiYan Hou, Xinyu Tang, Hongyan An, Jianjin Zhang, Weizhen Wang, Yunyun Han, Gengsheng Li, Xiangzhao Hao, Haiyun Guo, Wenbin Hu, Jinqiao Wang, Yafeng Deng · 7 August 2026

  Reinforcement learning with verifiable rewards (RLVR) improves the reasoning capabilities of large language models using automatically verifiable outcome signals, but these signals are typically sparse and at the sequence-level. On-policy self-distillation (OPSD) mitigates this sparsity by querying …

• Simulator-Grounded Large Language Models for Industrial Causal Reasoning: Tool-Use, Structured Injection, and Plant-Portable Retrieval for Wastewater Treatment Decision Support

  Gary Simethy, Daniel Ortiz Arroyo, Petar Durdevic · 7 August 2026

  Wastewater operators need answers grounded in how their plant's variables interact and how fast effects propagate, not in generic pretraining text, when asking causal questions such as "why is N2O rising?" or "what happens if I cut aeration by 20%?". We compare three concrete ways to ground a frozen…

• Hierarchical Latent Prediction for Language Models

  Chang Shi, Tim Pearce, Manan Tomar, Siddhartha Sen, John Langford · 7 August 2026

  While standard Next-Token Prediction (NTP) lays the foundation of language model pre- training, its teacher-forced training paradigm may not be optimal for long-horizon reasoning and planning. Recent works such as Multi-Token Prediction (MTP) and Next-Latent prediction (NextLat) try to mitigate the …

• Reasoning Errors Have a Region and a Direction in the Residual-Stream Trajectory of LLMs

  Hamed Damirchi, Ignacio Meza De la Jara, Damith Ranasinghe, Yuhang Liu, Javen Shi · 7 August 2026

  As language models are increasingly used for tasks that require verifiable reasoning, reliably distinguishing sound reasoning from flawed reasoning has become an important practical problem. Recent trajectory-based methods seek this signal in layerwise residual-stream displacements, which capture ho…

• Learning When to Trust via Selective Context Preference Optimization

  Xian Sun, Wei Chow, Yingshuo Wang, Junhao Liu, Wei Gao, Qing Wu, Lingdong Kong · 7 August 2026

  Language models increasingly condition their answers on external signals, and a single misleading one can turn a correct answer wrong. The obvious remedy, training models to resist such signals, hides a failure mode: a model that ignores all context looks robust yet is useless when the context is wo…

• Answer First, Reason Later: Commitment Order in Diffusion LLMs

  Jewon Yeom, Jaewon Sok, Seonghyeon Park, Jeongjae Park, Hwiyeong Lee, Taesup Kim · 7 August 2026

  Masked diffusion language models (dLLMs) can commit tokens in any order -- a freedom marketed as their core advantage over autoregressive decoding. We show that on reasoning tasks this freedom is instead the axis of failure. Logging every commitment during decoding of LLaDA-8B on GSM8K, we find that…

• Woodpecker Distillation: Weak Models Diagnose Reasoning Bugs in Strong Models

  Dayu Wang, Jiaye Yang, Weikang Li, Jiahui Liang, Yang Li, Deguo Xia, Jizhou Huang · 7 August 2026

  Large language models often fail on reasoning tasks despite possessing the capability to solve them. We argue that many such failures arise from localized reasoning bugs in intermediate steps rather than from global incompetence. We show that these bugs are frequently repairable: inserting a short p…

• Refining Over Resampling: Test-Time Self-Correction for LLM Reasoning

  Ahsan Bilal, Muhammad Ahmed Mohsin, Muhammad Umer, Lena Trigg, Ali Subhan, Muhammad Ali, Dean F. Hougen · 7 August 2026

  Test-time scaling improves LLM reasoning by using additional inference compute, but wider sampling alone can suffer from diminishing returns: new rollouts often repeat existing answer patterns instead of adding useful reasoning diversity. Verifier-based selection offers an alternative, but its perfo…

• Looking in the Mirror: Introspecting Side-Effect Misalignments Induced by Fine-Tuning

  Kotaro Yoshida, Laura Gomezjurado Gonzalez, Yukinori Yamamoto, Yuji Naraki, Ryotaro Shimizu, Wenya Wang · 6 August 2026

  Fine-tuning enables a source model to acquire desired capabilities and behaviors in a target domain while retaining much of its general-purpose competence. However, this adaptation process can also degrade alignment properties that were present in the source model. Recent work has shown that large l…

• Large-Small Model Collaboration for Enhancing Edge-Deployed Small Models

  Peigen Liu, Yijiang Fan, Zixuan Xu, Yuren Mao, Longbin Lai, Ying Zhang · 6 August 2026

  Edge devices host domain-specific small language models (SLMs) with limited resources, while private clouds offer larger LLMs. We propose G-Boost, an adaptive edge-cloud framework that improves a deployed SLM's task performance without parameter updates. It formulates reasoning as a tree search, cho…

• SpecRoll: Fast-Slow Verifier-Feedback Adaptation for Speculative Reinforcement Learning Rollouts

  Nhat Minh Pham, Duy Tung Doan, Thi Duyen Ngo, Vinh Van Nguyen, Khac-Hoai Nam Bui · 6 August 2026

  Reinforcement learning (RL) post-training improves the reasoning capabilities of large language models, but autoregressive rollout generation remains a major efficiency bottleneck. Speculative decoding can accelerate generation, yet applying it during RL is difficult because the target policy contin…

• Provable Limits and Certified Deferral for Verbalized Uncertainty in Small Language Models

  Jianru Shen · 6 August 2026

  Small open-weight language models increasingly run in private, offline, and cost-sensitive settings, where the key deployment question is not only what a model answers but when it should defer to a human. We study whether verbalized confidence can support risk-controlled deferral, evaluating eleven …

• Agentic Reinforcement Learning with Observation-Calibrated Self-Distillation

  Yi Yang, Cong Qin, Xiaodan Liu, Chishui Chen, Qing Dong, Yan Zhang, Cao Liu, Zhao Yang, Lu Pan, Jiaye Lin, Yi Feng · 6 August 2026

  Large language model agents are commonly trained through reinforcement learning with sparse trajectory-level rewards, which offer limited guidance on how strongly individual tokens should be updated. On-Policy Self-Distillation (OPSD) addresses this by re-scoring generated tokens under a privileged …

• Chained Recursive Language Models for Multi-Iteration Reasoning

  Purbesh Mitra, Sennur Ulukus · 6 August 2026

  Long context reasoning in large language models (LLMs) is usually constrained by the fact that a single inference trajectory has to simultaneously explore the context, store intermediate state, verify evidence, and produce the final answer. This becomes particularly difficult in tasks that require e…

• Eliciting Intrinsic Hallucinations in LLMs via Semantically Equivalent Adversarial Attacks

  Atri Vivek Sharma, Brian Formento, Alessio Lomuscio · 6 August 2026

  Large language models (LLMs) are often used in conjunction with external knowledge sources to improve their factual accuracy and decrease hallucinations, through methods such as Retrieval-Augmented Generation (RAG). However, these systems remain susceptible to intrinsic hallucinations, where the mod…

• D$^2$F-ReAG: Dynamic Decomposition and Filtering for Multi-Hop Reasoning-Augmented Generation

  Jiaoyang Li, Junhao Ruan, Shengwei Tang, Kaiyan Chang, Zhengtao Yu, Tong Xiao, Jingbo Zhu · 6 August 2026

  Large language models (LLMs) often generate inaccurate answers due to their reliance on static internal knowledge. Retrieval-augmented generation (RAG) addresses this limitation by integrating external knowledge and excelling at single-hop queries. However, it struggles with multi-hop questions that…

• Protoreasoning in Tiny Transformers

  Eduardo Valle, Fergal Reid · 6 August 2026

  We show that tiny transformers can profitably employ a simple form of Chain of Thought, which we call protoreasoning, allowing us to study step-by-step reasoning on ~1M-parameter models and opening up opportunities for much more detailed experimentation and analysis than is feasible for larger model…

• PURPOSE: Poisoning Conflict Resolution in RAG via Proxy-Fact-Grounded Updates

  Zijian Wang, Yubo Zhu, Muzhi Dong, Yanjun Lou, Yisheng Li, ZiLiang Zhang, Wei Tong, Yuan Zhang, Jingyu Hua, Sheng Zhong · 6 August 2026

  In Retrieval-Augmented Generation (RAG), post-retrieval conflict resolution arbitrates among noisy or contradictory retrieved passages. However, the robustness of this safeguard against knowledge poisoning has not been adequately studied. Existing black-box poisoning methods all assert the target an…

• MIDAS: Multi-LLM Iterative Data-Adaptive Summarization

  Karen Lee, Dhanashree Balaram, Seojun Shon, Umair Rasheed · 6 August 2026

  Text summarization is deceptively difficult. While condensing information seems straightforward, real-world enterprise summarization of support tickets, legal documents, incident reports, and more, demands strict adherence to domain-specific guidelines, output formats, and organizational conventions…